Aircraft Ice Accretion Prediction Using Neural Network and Wavelet Packet Transform

James Thompson, Hongwei Wu, S Chang, M Leng

Research output: Contribution to journalArticle

Abstract

Purpose – This paper presents a new technique based on the combination of wavelet packet transform (WPT) and artificial neural networks (ANNs) for predicting the ice accretion on the surface of an airfoil.

Design/methodology/approach – Wavelet packet decomposition is used to reduce the number of input vectors to ANN and improves the training convergence. An artificial neural network is developed with five variables (velocity, temperature, liquid water content, median volumetric diameter and exposure time) taken as input data and one dependent variable (the decomposed ice shape) given as the output. For the purpose of comparison, three different artificial neural networks; back-propagation network (BP), radial basis function network (RBF) and generalized regression neural network (GRNN), are respectively trained to simulate the wavelet packet coefficients as a function of the in-flight icing conditions.

Findings – The predicted ice accretion shapes are compared with the corresponding results from previously published NASA experimentation, LEWICE and the Fourier-expansion-based method. It is found that the BP network has an advantage on predicting the rime ice, and the RBF network is relatively suitable for the glaze ice, while the GRNN can be applied for both without classifying the specimens. The results also show an advantage of WPT in performing the analysis of ice accretion information and the prediction accuracy is improved as well.

Practical implications – The proposed method is open to further improvement and investment due to its small computational resource requirement and efficient performance.

Originality/value – The simulation method combining ANN and WPT outlined here can lay the foundation for further research relating to ice accretion prediction under different ice cloud conditions.
Original languageEnglish
Pages (from-to)128-136
JournalAircraft Engineering and Aerospace Technology
Volume88
Issue number1
DOIs
Publication statusPublished - 2016

Fingerprint

Ice
Aircraft
Mathematical transformations
Neural networks
Radial basis function networks
Backpropagation
Glazes
Airfoils
Water content
NASA
Decomposition
Liquids

Keywords

  • Aircraft icing
  • Neural network
  • Conformal mapping
  • Wavelet packet transform

Cite this

@article{647414c0726b4d59a704d9905b980618,
title = "Aircraft Ice Accretion Prediction Using Neural Network and Wavelet Packet Transform",
abstract = "Purpose – This paper presents a new technique based on the combination of wavelet packet transform (WPT) and artificial neural networks (ANNs) for predicting the ice accretion on the surface of an airfoil.Design/methodology/approach – Wavelet packet decomposition is used to reduce the number of input vectors to ANN and improves the training convergence. An artificial neural network is developed with five variables (velocity, temperature, liquid water content, median volumetric diameter and exposure time) taken as input data and one dependent variable (the decomposed ice shape) given as the output. For the purpose of comparison, three different artificial neural networks; back-propagation network (BP), radial basis function network (RBF) and generalized regression neural network (GRNN), are respectively trained to simulate the wavelet packet coefficients as a function of the in-flight icing conditions.Findings – The predicted ice accretion shapes are compared with the corresponding results from previously published NASA experimentation, LEWICE and the Fourier-expansion-based method. It is found that the BP network has an advantage on predicting the rime ice, and the RBF network is relatively suitable for the glaze ice, while the GRNN can be applied for both without classifying the specimens. The results also show an advantage of WPT in performing the analysis of ice accretion information and the prediction accuracy is improved as well.Practical implications – The proposed method is open to further improvement and investment due to its small computational resource requirement and efficient performance.Originality/value – The simulation method combining ANN and WPT outlined here can lay the foundation for further research relating to ice accretion prediction under different ice cloud conditions.",
keywords = "Aircraft icing, Neural network, Conformal mapping, Wavelet packet transform",
author = "James Thompson and Hongwei Wu and S Chang and M Leng",
year = "2016",
doi = "10.1108/AEAT-05-2014-0057",
language = "English",
volume = "88",
pages = "128--136",
journal = "Aircraft Engineering and Aerospace Technology",
issn = "1748-8842",
publisher = "Emerald Publishing Limited",
number = "1",

}

Aircraft Ice Accretion Prediction Using Neural Network and Wavelet Packet Transform. / Thompson, James; Wu, Hongwei; Chang, S; Leng, M.

In: Aircraft Engineering and Aerospace Technology, Vol. 88, No. 1, 2016, p. 128-136.

Research output: Contribution to journalArticle

TY - JOUR

T1 - Aircraft Ice Accretion Prediction Using Neural Network and Wavelet Packet Transform

AU - Thompson, James

AU - Wu, Hongwei

AU - Chang, S

AU - Leng, M

PY - 2016

Y1 - 2016

N2 - Purpose – This paper presents a new technique based on the combination of wavelet packet transform (WPT) and artificial neural networks (ANNs) for predicting the ice accretion on the surface of an airfoil.Design/methodology/approach – Wavelet packet decomposition is used to reduce the number of input vectors to ANN and improves the training convergence. An artificial neural network is developed with five variables (velocity, temperature, liquid water content, median volumetric diameter and exposure time) taken as input data and one dependent variable (the decomposed ice shape) given as the output. For the purpose of comparison, three different artificial neural networks; back-propagation network (BP), radial basis function network (RBF) and generalized regression neural network (GRNN), are respectively trained to simulate the wavelet packet coefficients as a function of the in-flight icing conditions.Findings – The predicted ice accretion shapes are compared with the corresponding results from previously published NASA experimentation, LEWICE and the Fourier-expansion-based method. It is found that the BP network has an advantage on predicting the rime ice, and the RBF network is relatively suitable for the glaze ice, while the GRNN can be applied for both without classifying the specimens. The results also show an advantage of WPT in performing the analysis of ice accretion information and the prediction accuracy is improved as well.Practical implications – The proposed method is open to further improvement and investment due to its small computational resource requirement and efficient performance.Originality/value – The simulation method combining ANN and WPT outlined here can lay the foundation for further research relating to ice accretion prediction under different ice cloud conditions.

AB - Purpose – This paper presents a new technique based on the combination of wavelet packet transform (WPT) and artificial neural networks (ANNs) for predicting the ice accretion on the surface of an airfoil.Design/methodology/approach – Wavelet packet decomposition is used to reduce the number of input vectors to ANN and improves the training convergence. An artificial neural network is developed with five variables (velocity, temperature, liquid water content, median volumetric diameter and exposure time) taken as input data and one dependent variable (the decomposed ice shape) given as the output. For the purpose of comparison, three different artificial neural networks; back-propagation network (BP), radial basis function network (RBF) and generalized regression neural network (GRNN), are respectively trained to simulate the wavelet packet coefficients as a function of the in-flight icing conditions.Findings – The predicted ice accretion shapes are compared with the corresponding results from previously published NASA experimentation, LEWICE and the Fourier-expansion-based method. It is found that the BP network has an advantage on predicting the rime ice, and the RBF network is relatively suitable for the glaze ice, while the GRNN can be applied for both without classifying the specimens. The results also show an advantage of WPT in performing the analysis of ice accretion information and the prediction accuracy is improved as well.Practical implications – The proposed method is open to further improvement and investment due to its small computational resource requirement and efficient performance.Originality/value – The simulation method combining ANN and WPT outlined here can lay the foundation for further research relating to ice accretion prediction under different ice cloud conditions.

KW - Aircraft icing

KW - Neural network

KW - Conformal mapping

KW - Wavelet packet transform

U2 - 10.1108/AEAT-05-2014-0057

DO - 10.1108/AEAT-05-2014-0057

M3 - Article

VL - 88

SP - 128

EP - 136

JO - Aircraft Engineering and Aerospace Technology

JF - Aircraft Engineering and Aerospace Technology

SN - 1748-8842

IS - 1

ER -